CHAPTER 3:

INTRODUCTION TO THE NOx ABATEMENT PROCESS AT

RFAAP AND COMPUTER SIMULATION OF THE NOx

ABATEMENT PROCESS AS A CONVERSION-REACTION MODEL

61
3.1 Introduction

3.1.1 An Overview

This chapter introduces the reader to a specific application, or case study, of NOx

removal technology, that being the NOx abatement system installed at the Radford

Facility and Army Ammunition Plant (RFAAP), Radford, Virginia. The system at

RFAAP reduces NOx emissions from fumes emanating from the cellulose nitration

reactors at the site. Two major components make up the continuous NOx abatement

system. A scrubber/absorber tower contacts the NOx-laden fumes with water, producing

nitric acid from the aqueous absorption of NOx gases. Downstream of the

scrubber/absorber tower, a reaction vessel with a honeycomb-structured heterogeneous

catalyst bed reduces NOx with ammonia to atmospheric nitrogen and water. We give a

brief summary of the reaction chemistry occurring in these two units as proposed by

Radford.

Further, we introduce the trial-and-error method by which we plan to achieve an

accurate working computer model for exploring system behavior and design

optimization. We present an ASPEN Plus computer simulation for the NOx abatement

data using stoichiometric conversion reactor models to approximate the chemistry

involved. We show that the model is capable of accurately simulating the base-case data.

At this point, we have simply calculated a mass balance and hypothesized the

62
predominant reaction pathways. We note and explain the discrepancies between our

findings and the material balance provided by personnel at RFAAP.

When we attempt to extend the model to various design and process variables

adjustments through sensitivity tests, the model is found lacking. It cannot represent the

behavior shown in the experimental literature for similar processes. Thus, more work

needs to be done on the model and aspects of the reaction chemistry used. Though the

chemistry utilized in the model represents a “stoichiometrically accurate” approximation

(it can fit the base-case mass balance), it fails mechanistically and does not follow true

system behavior in response to manipulated variables.

3.1.2 Introduction to NOx Removal at RFAAP

The literature review in Chapter 2 concerns itself primarily with an overview of

the most utilized NOx removal technology and a brief introduction to absorption coupled

with SCR. The remainder of this document focuses on the process located at the Radford

Facility and Army Ammunition Plant (RFAAP) in Radford Virginia. We hope

applications from this thesis can be extended to similar processes in operation, in

construction, and in the design stages at other sites. The process at RFAAP makes use of

absorption/SCR technology as discussed in Section 2.4.

In the sections on computer model development, we state all assumptions and

address their consequences on the generalization of results and conclusions to similar

63
processes. Conclusions referring to simulation results apply to the range of operating

conditions explored here. Similar processes under different conditions can make certain

of our assumptions invalid for those processes in question. The complexity of NOx

chemistry and processing makes proper selection of assumptions and operating

conditions of utmost importance.

3.1.3 Purpose of NOx Abatement at RFAAP

The nitrocellulose plant at RFAAP uses highly concentrated nitric and sulfuric

acid to nitrate cotton linters to produce nitrocellulose. The reaction is run at atmospheric

pressure in large stainless steel vats (atmospheric pressure for the altitude of the Radford

site is approximately 14.1 psia, 0.959 atm, 97.2 kPa). Concentrated nitric acid can

vaporize and also decompose in water to form NOx gases in appreciable amounts.

Sulfuric acid can do the same to a lesser extent. Without some form of treatment for

these fumes, they escape to the atmosphere causing damage to the environment and

subject RFAAP to fines levied by governmental agencies.

In 1993, RFAAP purchased and installed a NOx abatement system for removing

nitrogen oxide (NOx) gases from fumes that originate from their nitrocellulose plant.

The NOx abatement system performs flue-gas treatment essential to the compliance of

the plant to new guidelines laid down by the Environmental Protection Agency’s (EPA)

Clean Air Act and further restrictions outlined by the Commonwealth of Virginia

Department of Environmental Quality (DEQ). The emissions limits set aside for the

64
effluent fumes from the nitrocellulose line at RFAAP in the stationary source permit from

DEQ are as follows: 125 ppmv (as an hourly average), 2.8 lbs/hr, or 12.3 tons/yr for

NOx. Sulfur dioxide limits are as follows: 92 ppmv (as an hourly average), 4.0 lbs/hr, or

17.5 tons/yr.

The recent performance of the SCR has not been satisfactory. Considering its

intermittent shutdown, tightening environmental regulations, and the possibility of

optimization and retrofit options, we choose the NOx abatement system and all

associated equipment for ASPEN Plus computer simulation and modeling.

3.1.4 Process Description

In brief, the process receives NOx-rich fumes (NOx, in this case, consists of

mostly NO2 and NO with other NOx compounds formed from the dimerization of these

two molecules, with NO2 making up the bulk) from the nitrocellulose production line on

site. As alluded to in Chapter 2, the situation where NO and NO2 are mixed in the vapor

provides the engineer with unique challenges, as well as opportunities for treatment. In

general, removing NO and NO2 require different processes to be successful. RFAAP

utilizes absorption to remove NO2 and SCR to remove NO.

A scrubber/absorber section absorbs some of the NOx gases and converts a

portion to nitric acid. The remaining NOx fumes proceed to a catalyst vessel that uses

ammonia (fed by steam vaporization) to reduce the NOx to atmospheric nitrogen, N2.

65
The gases proceed from the catalyst vessel to the stack and out to the atmosphere. Figure

3.1 shows the simplified NOx abatement process.

66
 Ammonia Steam

Filtered Water

Scrubber/ Catalyst Stack Gas

Absorber Vessel
NOx Fumes

Dilute Nitric Acid

Figure 3.1. Simplified design of the NOx abatement process flow diagram.

67
 We show a highly simplified block flow diagram for NOx abatement in Figure

3.1. This figure clearly shows the two main sections of the process. The

scrubber/absorber mainly removes NO2 and its derivatives (actually producing a limited

amount of NO), whereas the catalyst vessel mainly removes NO. The two sections

perform different tasks in the reduction of NOx and also work relatively independently of

each other. We will treat them as such in the computer simulation discussion.

A scrubber/absorber unit, a stream preheater, a feed/effluent heat exchanger, a

direct-fired natural gas heater, and a catalyst vessel reactor make up the main components

of the process. These units can be seen in the more detailed block flow diagram

represented in Figure 3.2. Fumes from the nitrocellulose (NC) line (greater than 97% air

by volume with the remainder being NOx gases) feed into the bottom of the

scrubber/absorber. Filtered water feeds to the top of the scrubber/absorber. The

scrubber/absorber forces contact between the fumes and the water in order to absorb as

much of the NOx gases as possible into the aqueous phase. Absorbed NOx compounds

react with water to form nitric acid. The absorbing section separates the dilute nitric acid

from the NOx gases. Dilute nitric acid leaves the bottom of the scrubber for waste-acid

treatment or recovery.

The remaining NOx gases leave the top of the absorber for further treatment by

selective catalytic reduction (SCR). A demister removes any entrained water vapor and

nitric acid. A preheater in the line heats the NOx gases with steam. A feed/effluent heat

exchanger further heats the NOx gases using the outlet from the catalyst vessel as the hot

68
stream. From there, a direct-fired heater uses natural gas combustion to heat the NOx

gases to the required reaction temperature.

Ammonia feeds just ahead of the catalyst vessel, mixing with the NOx fumes.

Automated control of the ammonia feed ensures proper reaction stoichiometry to

minimize the ammonia slip, that is, unreacted ammonia that leaves the catalyst vessel in

the effluent, and to maximize the amount of NOx reduced. Steam also feeds to the

catalyst vessel, serving to diffuse and vaporize the ammonia in a safe manner. The

catalyst promotes several simultaneous reactions, which will be discussed in detail later,

that reduce the NOx gases to N2 (atmospheric nitrogen) and H2O with ammonia as the

reducing agent. The outlet stream of the catalyst vessel preheats the NOx gas in the

aforementioned feed/effluent heat exchanger. The catalyst-vessel effluent passes through

the feed/effluent heat exchanger on its way to the vent stack. Figure 3.2 shows the

arrangement of equipment and streamlines in the NOx abatement system.

69
 Demister

Filtered Water
Steam Heater

Ammonia

Scrubber/Absorber
Steam

To Stack

Heat Exchanger
Feed-Effluent
Catalyst Vessel

Bottom
Recycle Direct-Fired Heater
NOx-Rich
Fumes

Natural
Exhaust

Gas
Bottoms
To Waste Acid
Treatment

Figure 3.2. Block flow diagram of NOx abatement system provided by RFAAP.

70
3.2 Detailed Explanation of Process Equipment

3.2.1 Scrubber/Absorber

The scrubber/absorber is the first major unit in the NOx abatement system. The

introduction briefly outlined the purpose of this unit. A more detailed examination

follows. The column consists of two sections: The top of the column is the staged

absorber section, and the bottom is a two-stage scrubber tank with its familiar spray

nozzles. Fumes composed of air rich in NOx gases from the nitrocellulose line of

RFAAP feed into the bottom of the scrubber section, while filtered water feeds into the

top of the absorber section. Filtered water covers the absorber stages and flows down the

downcomers. The staged absorber section consists of 16 bubble-cap trays. Each tray has

79 equally spaced bubble-caps. Figure 3.3 details the upper portion of the

scrubber/absorber column. We number the absorption section bubble-cap trays from the

top down as per the ASPEN convention. Figure 3.4 shows the tray arrangement. Figure

3.5 gives an expanded view of a bubble cap and downcomer area of a tray.

71
D owncomer Area D owncomer Area

Bubble Caps

Figure 3.3. Schematic of bubble-cap tray arrangement.

72
 Liquid Flow
Liquid Level

Cap Weir

Tray Deck
Gas Bubbles
Holes in Cap
Gas Flow for Gas Flow

Figure 3.4. Cutaway view bubble cap and tray.

73
 Filtered Water Fumes to Demister
and on to Further
Treatment with SCR
1
2
3
4
Tray 5
Numbers 6
7
8
9
10
11 Tray Spacing
Weir Height
12
13
14
15
16
To Scrubber
Section

Figure 3.5. Detailed schematic of absorption section of scrubber/absorber tower.

74
 The scrubber section begins below the bottom tray of the absorber. Figure 3.6

details the scrubber section of the scrubber/absorber tower. The scrubber section consists

of two spray nozzle stages. Figure 3.7 shows the design of the spray nozzles within the

scrubber section of the column. A catch basin collects the water and funnels it to a

recycle line. A pump in that line pumps the water up to a spray-nozzle section just below

the bottom absorber tray. Water that overflows this catch basin goes to the bottom of the

column where it is drained. Pumps send it to a splitter where some is recycled back to a

second spray nozzle below the first nozzle. The remainder leaves as a waste-acid stream.

A vent-gas fan blows the fumes from the NC line to the bottom of the column. The

fumes contact the sprayed water in the scrubber section and bubble up through the

absorber trays. Some of the NOx gases are absorbed into the aqueous phase.

Reactions, which produce nitric acid from the NOx gases, occur throughout the

scrubber/absorber. Table 3.1 shows the reactions that occur in the scrubber/absorber as

proposed by personnel at RFAAP. In Chapter 4, we will show that this set of reactions,

though accurate as far as stoichiometry, is incomplete and misleading. However, correct

stoichiometry is all that is needed for a conversion-reaction model. Thus, we will

proceed.

75
 Bottom
Absorber Tray
Scrubber Acid Recycle Spray Nozzles (Top Scrubber)

Catch Basins
Chimney Plate

Scrubber Spray Nozzles (Bottom Scrubber)

Acid Pump To Waste
Bottom Acid Treatment
Acid
Recycle
Fumes from
Nitrocellulose
Line

Vent-Gas Fan Scrubber

Acid Pumps

Figure 3.6. Cutaway view of the scrubber section of the scrubber/absorber tower.

76
 Spray Nozzles Column Shell

Filtered
Water Inlet

Figure 3.7. Detailed view of the spray nozzles used in the scrubber section of the scrubber/absorber tower.

77
 Table 3.1. Reactions occurring in the scrubber/absorber as proposed by personnel

at RFAAP.

Rxn. # Reaction Stoichiometry

3.1 2NO + O2 ↔ 2NO2

3.2 2NO2 + H2O ↔ HNO3 + HNO2

3.3 3HNO2 ↔ HNO3 + H2O + 2NO

3.4 3NO2 + H2O ↔ 2HNO3 + NO

We see that reaction (3.4) results from the addition of reaction (3.3) and reaction

(3.2) multiplied by a factor of three. Additionally, nitric acid (HNO3) dissociates in

aqueous solution in an ionization reaction. Table 3.2 shows this reaction. Nitric acid is a

strong acid that quickly and almost completely dissociates in solution. Therefore,

ionization consumes HNO3 almost as quickly as it is produced. This tendency drives the

equilibrium of the above reactions to the products. Table 3.1 shows that the net result of

these reactions is the conversion of NO to NO2, and the subsequent conversion of NO2 to

HNO3 (nitric acid). Nitric acid then dissociates, yielding a proton (H3O+ is a more

realistic representation for than H+) and a nitrate ion as per reaction (3.5) in Table 3.2.

Figure 3.8 gives a visual representation of the reaction network for NOx absorption.

Table 3.2. Dissociation reaction of nitric acid in aqueous solution.

Rxn. # Reaction Stoichiometry

3.5 HNO3 + H2O ↔ H3O+ + NO3-

78
2NO + O2 ↔ 2NO2 2NO2 + H2O ↔ HNO3 + HNO2

x 3/2 x3
3NO2 + H2O ↔ 2HNO3 + NO 3HNO2 ↔ HNO3 + H2O + 2NO

x 1/2

HNO3 + H2O ↔ H3O+ + NO3-

Figure 3.8. Visual representation of reaction network proposed by RFAAP.

79
 This reaction set is not a pure chain reaction because a mixture of NO and NO2

are fed to the system. If NO was the main component of the NOx, as is the case in most

combustion processes, then reaction (3.1) would represent the rate-limiting reaction for

this sequence.

These reactions accomplish the dual goals of eliminating the source of NOx

emissions, as well as producing nitric acid (HNO3). Nitric acid, a compound that absorbs

into the aqueous phase readily and, after dissociation, has a strong tendency to stay there,

also represents a potentially valued product. In high enough concentrations, the nitric

acid can be recovered, not only adding to the product yield, but also eliminating the cost

required for wastewater treatment from this process.

Given a large enough residence time, the strong equilibrium tendency to products

for this reaction set facilitates pollutant removal. The production of the nonvolatile

nitrate ion (NO3-) and the insoluble NO as the final products of the reaction network

constitute the major driving forces for NO2 absorption. The nitric acid leaving the

scrubber/absorber is dilute and of a small quantity. It does not contribute greatly to the

wastewater duties of the plant. We will discuss the possibility of its recovery in later

sections. Recall that aside from the obvious nitric acid dissociation reaction, RFAAP

proposed this reaction network. Though we contend that it is incomplete as a

mechanism, the overall stoichiometry is correct, and we will attempt to model the process

using it. The reader should note that no distinction has been made for the reactions or the

80
components themselves as to which phase they take place or predominate. We will show

in this and the following chapters that this reaction heterogeneity becomes a major

problem for proper simulation. Thus, the eventual failure of the conversion model will

become the impetus for the subsequent equilibrium and kinetic models.

3.2.2 Demister

The demister, installed on the outlet of the top of the absorber column, consists of

a simple tank packed with wire-mesh screens. Entrained liquid impacts with the

screening and trickles to the bottom of the demister where it flows down to a drain that

leaves the wastewater treatment area of RFAAP. Figure 3.9 shows a simple schematic

diagram of the demister. No data are available from RFAAP on the composition or flow

rate of the liquid flow from the bottom of the demister aside from the fact that it

constitutes a “trickle.” Also, coming from the top stage of the absorber, this stream

should contain very little NOx or nitric acid. We have therefore assumed it to be

negligible.

81
 Cleanout
Flange

Vapor Out

Vapor and
Entrained Liquid
Impingement
Screening

Liquid to Drain

Figure 3.9. Detailed diagram of demister construction.

82
 3.2.3 Heat Exchangers and Process Heaters

The fumes leave the absorber at nearly ambient (approximately 80 oF, 27 oC)

temperature. For selective catalytic reduction with ammonia, the temperature of the

fumes must be raised to approximately 600 oF (316 oC), thus requiring a temperature rise

of at least 520 oF (289 oC) for a flow rate of approximately 18,600 lb/hr (4000 scfm)

consisting mostly of air. Table 3.3 lists the different temperature-change operations in

the NOx abatement system, including the flowsheet connectivity and inlet/outlet

temperatures.

83
 Table 3.3. Process streams, heating units and the temperature changes affected.

Equipment Name Stream In From: Stream Out Inlet Temp., Outlet Temp.,
o
To: F (oC) o
F (oC)

Preheater Demister Economizer 80 (27) 100 (38)

Economizer Preheater Direct-Fired Heater 100 (38) 406 (208)

Economizer Catalyst vessel Stack 650 (343) 350 (177)

Direct-Fired Heater Economizer Catalyst Vessel 357 (181) 600 (316)

Notes: values are those provided by Radford

The preheater uses utility steam to raise the fume temperature from 80 oF (27 oC)

to 100 oF (38 oC). After the preheater, the countercurrent economizer (also known as the

feed/effluent heat exchanger) uses the effluent from the catalyst vessel to raise the fume

temperature to 406 oF (208 oC). The direct-fired heater combusts natural gas within the

shell, while the fumes flow through a U-tube in the heat exchanger. Within this unit, the

fume temperature rises to approximately 600 oF (316 oC). Selective catalytic reduction is

highly exothermic so the fume temperature rises within the catalyst vessel, from 600 oF

(316 oC) to 650 oF (343 oC). Figures 3.10, 3.11, and 3.12 show the steam preheater,

feed/effluent heat exchanger, and direct-fired heater respectively.

84
Fumes In Steam In
80 oF 260 psig

Condensate Out Fumes Out

100 oF

Figure 3.10. Detailed diagram of steam preheater.

85
 Honeycomb Catalyst Modules

Fumes from
Steam Preheater

100 oF

Fumes from Fumes

Catalyst Vessel to Stack

650 oF 350 oF

Fumes to
Direct-Fired Heater

Catalyst Vessel 300 oF

Figure 3.11. Detailed diagram of feed/effluent heat exchanger (economizer).

86
 Combustion Products to Stack

Fumes In
300 oF Fumes Out
Tubes 600 oF

Combustion

Natural Gas or #2 Fuel Oil Combustion Air

Figure 3.12. Detailed diagram of direct-fired heater.

87
 3.2.4 Catalyst Vessel

Ammonia and steam mix with the NOx gases downstream of the direct-fired

heater. This mixture feeds into the catalyst vessel. Ammonia reduces the NOx gases

over a vanadia-supported-on-titania (V2O5 on TiO2) catalyst bed. Catalyst blend recipes

are highly proprietary and we were not given exact details on the makeup of the catalyst.

We know that it consists of less than 5% vanadia on less than 85% titania (both weight

percentages). Reaction rate and selectivity depends heavily upon catalyst blend recipe.

However, the mass balance provides a good starting point. Table 3.4 shows the reactions

that are thought to occur in the catalyst vessel.

Table 3.4. Reactions proposed for the catalyst vessel.

Rxn. # Reaction Stoichiometry

3.6 4NH3 + 6NO ↔ 5N2 + 6H2O

3.7 4NH3 + 4NO + O2 ↔ 4N2 + 6H2O

3.8 2NO2 + 4NH3 + O2 ↔ 3N2 + 6H2O

3.9 NO + NO2 + 2NH3 ↔ 2N2 + 3H2O

Note: These reactions are provided by RFAAP

We see that the net effect of the selective catalytic reduction reaction is to

consume NH3 and NOx gases, while producing water and atmospheric nitrogen. We do

not know which reactions predominate. However, at the point of input into the catalyst

vessel, NO represents the preponderance of the NOx.

88
 The catalyst vessel itself consists of stainless steel construction roughly in a three-

dimensional rectangular prism configuration. Within this large rectangular prism are

smaller boxes, called modules, that house the honeycomb catalyst bundles, called

elements. These honeycomb catalyst elements, consist of support and active catalyst

materials in a single piece. Hollow rectangular holes, called cells, run the length of the

element and provide for high gas flow rates, as well as a large surface-area-to-volume

ratio. Each element has a 20-by-20 cell arrangement for 400 total cells per element (cell

number depends on application). Figure 3.13 illustrates a catalyst element used at

RFAAP.

As they consist of fragile perforated ceramic, catalyst elements require a steel

superstructure for transport and installation to reduce the chance of breakage. Many

elements are bundled together within this steel structure to make up a module. We refer

to the steel superstructure around the bundle of elements as the frame. The surface of the

frame also interlocks with other framed modules acting as a gasket to seal the gas flow

upon installation. Figure 3.14 shows a typical frame for a horizontal gas-flow application

(similar to that used at RFAAP). Figure 3.15 gives a detailed view of the arrangement of

the catalyst elements within the catalyst-module frame. Figure 3.16 shows the

arrangement of the modules. The catalyst vessel at RFAAP runs at atmospheric pressure

requiring only light construction. Figure 3.17 illustrates the construction of the selective

catalytic reduction vessel which houses eight catalyst modules.

89
Figure 3.13. Detailed diagram of honeycomb catalyst element.

90
Figure 3.14. Catalyst module for horizontal flow, similar to those used by RFAAP.

91
 Steel Frame

Catalyst Element

Figure 3.15. Detailed view of catalyst module with exploded view of catalyst element.

92
 Flow
s
Ga

Figure 3.16. Catalylst module arrangement within the catalyst vessel.

93
 Ammonia

Honeycomb Catalyst Modules

Steam

Fumes
to Stack
Fumes from
Direct-Fired
Heater

600 oF
Feed/Effluent
Heat Exchanger
Catalyst Vessel

Figure 3.17. Diagram of catalyst vessel in relation to material streams and adjacent equipment.

94
 3.2.5 NOx Abatement System Input and Output Flow Rates

Table 3.5 shows the input streams to the system as provided by RFAAP. Table

3.6 shows the output streams as provided by RFAAP. There are, however, more inputs

and outputs than specified for us by RFAAP. In some cases, ASPEN helps us determine

what those flows consist of. Other cases require us to make certain assumptions about

the streams. Simple component mass balances fill in any holes left in the information.

Table 3.7 shows the missing streams and their property values. The values have been

calculated based on assumptions we make from information provided by RFAAP. The

demister drain stream has been assumed to be negligible. RFAAP give the steam flow

rate to the catalyst vessel as 21 lb/hr at 450 oF (232 oC). We assume 100% quality steam

(all vapor) to arrive at the steam pressure of 422 psia (28.72 atm, 2910 kPa).

95
 Table 3.5. Input flows to the system (values given by RFAAP).

Stream: Fumes from Filtered Steam to Ammonia to

NC Line Water SCR SCR

Mole Flow

lbmol/hr

NO 0.39

NO2 3.37

H3N 1.37

O2 132.2

N2 497.3

H2O 15.1 22.17 1.17

H2SO4

HNO2

HNO3

Total Flow lbmol/hr 648.34 22.2 1.17 1.37

Total Flow lb/hr 18601 399 21 23.3

Total Flow cuft/hr 75600 6.42 24.3 264

Temperature F 90 78 450 86

Pressure psi 20 44 422 30

96
 Table 3.6. Output flows from the system (values given by RFAAP).

Stream: To Waste Acid Fumes to

Treatment Stack

Mole Flow

lbmol/hr

NO 0.06

NO2 Trace

H3N Trace

O2 132.06

N2 496.77

H2O 13.1 23.86

H2SO4

HNO2

HNO3 2.1

Total Flow 20.8 654.75

lbmol/hr

Total Flow lb/hr 376 18631

Total Flow cuft/hr 0.094 4000

Temperature oF 86 350

Pressure psi 15.1 14.1

97
 Table 3.7. Streams not specified by RFAAP (calculated or assumed).

Stream: Steam to Catalyst Demister Drainage

Vessel

Input Output

Mole Flow

lbmol/hr

NO

NO2

H3N

O2

N2

H2O 1.17

H2SO4

HNO2

HNO3

Total Flow 1.17 0

lbmol/hr

Total Flow lb/hr 21 0

Total Flow cuft/hr 24.3 0

Temperature oF 450 80

Pressure psi 422 14.1

98
3.3 Introduction to ASPEN Plus Simulation of NOx Abatement

3.3.1 Purpose for ASPEN Plus Simulation of NOx Abatement

The ideal mathematical/computer model is the simplest, most theoretically elegant

model that accurately simulates the actual process in its base configuration. The model

involves all the assumptions made about the process. First, the model must accurately

reflect the base-case data for the actual process; this will be referred to as model

accuracy. Next, it must account for results of process changes observed for the actual

process; this will be referred to as model precision. We refer to a simulation as when the

model is given inputs that are used to calculate outputs.

To determine the accuracy and precision of the model, we must give it known

inputs and run it as a simulation, comparing the results to known outputs. Achieving this

end requires a trial-and-error process that involves running the simulation, comparing the

results to plant data, and making changes to the model to affect its accuracy. In this

thesis, we will omit the details of how we arrive at each model in succession and focus on

the results and conclusions drawn from the simulations of the final evolution of each

model. Suffice it to say that much research and many simulations are run to arrive at the

logic and science of each model.

In any case, the engineer must thoroughly research the process to know how the

simulation should behave in response to process changes. The simulation must prove

99
itself to be accurate and precise before the engineer is justified in putting his/her faith in

it. After the model consistently proves its accuracy and precision, the engineer can

confidently approach previously unexplored process modifications, for which no previous

data exist. Even then, the model must continuously prove its worth to retain the

engineer’s confidence in the simulation results.

Quite often, a good model makes apparent conclusions about the physical system

that are not intuitively obvious. Filling in the holes in component mass and energy

balances, as we discussed in the previous section for this case study, represents a good

example of this ability of computer model simulations. We can use such conclusions as

stepping stones to further optimization.

3.3.2 Motivation for a Conversion-Reaction Model

The majority of unit operations encountered by chemical engineers involve heat

and mass transfer. It is often satisfactory to model these operations separately from

simultaneous reactions. Equilibrium assumptions usually model heat- and mass-transfer

processes quite accurately. Because most industrial reactions occur far from equilibrium,

when dealing with reactions, engineers often use a constant-percentage conversion with

respect to a key reactant to model a reactor.

The treatment of different processes separately often yields satisfactory results

because time, space, and equipment boundaries usually keep the separate operations of

100
heat and mass transfer and reaction relatively isolated from each other. Also, because of

the complexity of reaction engineering and the narrow range of acceptable operating

parameters, process and design engineers usually do not wish to affect the performance of

a working reactor. They may choose to achieve optimization in the preparation and

separation processes, while leaving reactors alone. For optimization of most industrial

reactors, the sheer complexity usually requires more research assets than are available to

the typical process engineer.

The particular case of NOx abatement at RFAAP includes a heterogeneous

catalytic reactor typical to many chemical plants. Here, we refer to the selective catalytic

reduction catalyst vessel. As described above, we are limited in our options for

improving the reactor performance because of the narrow range of performance for the

catalyst as well as other constraints on the SCR that we will discuss later.

In addition to the familiar plug-flow reactor, the process being studied also

includes a highly atypical absorption operation. In the scrubber/absorber, heat and mass

transfer are intimately coupled with chemical reaction. In both obvious and subtle ways,

temperatures, concentrations, and reaction rates and conversions affect each other. With

the conversion model and computer simulation, we shall see if percentage conversions

can accurately model the reactions in this operation under changes in process variables.

Also, we will investigate system sensitivity to certain process variables. The reaction

conversions will remain constant, but the thermal and vapor-liquid equilibrium will be

101
affected. We will compare the results to those found in the literature to determine the

appropriateness of the conversion model over the relevant range of conditions.

3.3.3 ASPEN Simulation Procedure

Computer modeling and simulation represents a powerful technique for the

process engineer. Equally useful for design and retrofit, the computer provides the

process engineer with calculations for predicting equipment and process behavior. The

speed at which information is available to the engineer by these techniques represents its

real usefulness. This speed also provides the engineer with the ability to produce large

quantities of erroneous data.

Computer simulations do not excuse the engineer from making critical

assumptions and properly specifying the system, or of digesting the information that is

produced. Modeling software is simply a black box that processes the engineer’s input

specifications and returns an output. Even with the proper input, the output still requires

the proper interpretation. With these tools, the engineer’s responsibility has not changed.

Perhaps the software itself poses the greatest challenge to the simulation engineer.

Chemical process simulation software makes implicit and explicit assumptions within the

unit-operation blocks that perform the calculations. The engineer must take these

assumptions and their ramifications into account when extending simulation results to

conclusions about the real process. Even with that caveat, computer simulation

102
represents a great improvement over attempts to apply literature studies to processes with

different equipment and operating conditions.

3.3.4 The Conversion-Reaction Model

Due to the vast array of chemical processes available to the chemical industry, no

single computer simulation block can accurately cover all possible equipment

configurations. However, combinations of computer-generated processes can adequately

model complex chemical behavior. The simulation engineer must configure the ASPEN

model to accurately simulate the behavior of the real process. Key differences can be

seen in the process as modeled by ASPEN when using stoichiometric conversion reactors

to model the chemical reactions. Figure 3.18 shows the process as modeled in ASPEN

for comparison.

The most readily apparent changes from the original process are the splitting of

the scrubber/absorber into an absorption column and the addition of two scrubber tanks,

as well as the addition of three stoichiometric reactors. Also, we add a flash unit below

the bottom scrubber tank. Each stream and simulation block is designated with an eight-

letter-maximum name. Figure 3.18, the detailed block flow diagram for the computer

model, shows all the stream and block designations. Stream and block names appear in

all capital letters.

103
 Demister PREHEAT
Absorber
ABSOUT FLASHOUT VAPOR
Filtered Water FILTE-W

DRAIN
To Drain HEATX
REACTOUT REACTOR

ABSREACT Steam STEAM

ABSLIQID
RECOUT To
SCRUBGAS PRODUCT
AMMONIA Stack
Ammonia
SCRUBTOP VAPOROUT
SCRUBBOT

SCRUBVAP SCRREACT HEATOUT

BOTWATER
Fumes From FUMES SCRUBBOT
NC Line DIRHEAT
BOTOUT
NO BOTRCYCL
To Waste Acid
BOT WEAKACID WASTACID Treatment
BOTREACT FLASHBOT SPLITTER

Figure 3.18. Flowsheet used for the conversion-model simulations.

104
 The unit-operations modules available in ASPEN cannot accurately model certain

real units that combine many chemical processes. As detailed earlier, the

scrubber/absorber column performs the tasks of a scrubber, an absorber, and a reactor.

Modeling this process accurately in ASPEN with a single tower would be difficult (if not

impossible). This is not generally a problem because, as mentioned before, we can

separate the different processes into individual ASPEN blocks and achieve satisfactory

results. This procedure also gives the added benefit of independent inspection of each

separate process individually. We can isolate problems and interesting results in

individual processes much easier this way. We can come away from the simulation with

more information because of this, although it remains under the caveat that the processes

had to be separated. This separate treatment may or may not introduce error in the

calculations.

We represent the scrubber/absorber (and reactor) as multiple blocks in ASPEN.

At the bottom, we use two scrubber tanks (two-phase flash drums) to represent the

scrubber section. At the top, we use a reactive absorption column to simulate the staged

absorber. We incorporate the reactions in four stoichiometric reactors because reactions

in the real process most likely run the whole height of the tower. We developed the

arrangement of stoichiometric reactors within the scrubber/absorber by trial and error,

comparing simulation results to RFAAP mass-balance data.

105
 First, a reactor is placed in the liquid exiting the bottom scrubber tank

(BOTREACT). This reactor simulates reactions occurring in the scrubber basin. A flash

drum, downstream of this reactor, releases the vapors formed by the reactions back up to

the scrubber (FLASHBOT). A second reactor is located in the liquid stream between the

upper and lower scrubber tanks (SCRREACT). Thirdly, a reactor placed in the vapor

stream between the top scrubber and absorber accounts for the initial contact between the

fumes and the sprayed water (ABSREACT). Fourthly, a reactive absorption model

simulates reactions on the absorber tower trays (ABSORBER). Table 3.8 summarizes the

units we add to the ASPEN Plus simulation. These units account for behavior that the

model units of the existing equipment could not.

Table 3.8. List of units added to ASPEN Plus simulation.

Unit (Block) Name Purpose

ABSREACT Gas-phase stoichiometric reactor

SCRREACT Liquid-phase stoichiometric reactor

SCRUBTOP Flash drum representing top scrubber stage

SCRUBBOT Flash drum representing top scrubber stage

BOTREACT Liquid-phase stoichiometric reactor representing scrubber basin

FLASHBOT Flash drum representing scrubber basin (releases NO)

We must mention absorber stage efficiencies at this point. Stage efficiency refers

to the degree to which vapor-liquid equilibrium is achieved on a stage. If we assumed

106
100% absorber stage efficiency, then our ASPEN Plus model of the absorber would have

16 trays, the same as the actual process. ASPEN Plus calculates absorber stgages as

equilibrium stages. These are, by definition, 100% efficient. We must account for the

inefficiency of real equipment, that is to say, the inability of most real processes to reach

complete equilibrium. The typical empirical value for absorption tower efficiency is

approximately forty percent (40%). Forty percent of 16 stages gives 6.4 theoretical

stages. We use 7 stages for our ASPEN Plus model of the absorption tower. This

number of theoretical trays translates to a tower efficiency of 44%.

ASPEN calculates the equilibrium for the HNO3 dissociation reaction using the

Electrolyte NRTL thermodynamic model. After specifying the components, ASPEN

calculates the equilibrium for the HNO3 dissociation reaction globally (in all blocks and

streams). However, if we do not explicitly specify the dissociation reaction for the

reactive-distillation column, ASPEN assumes the reaction does not occur in the column.

ASPEN assumes this because all reactions in the column must be specified explicitly.

This logic conflicts with and overrides the global specification of the reaction. Therefore,

reaction (3.5), the HNO3 dissociation reaction, must be specified in the ASPEN Plus

reaction sheet along with the other reactions in the column,. Exemption of this reaction

hinders the simulated performance of the column dramatically.

Primarily, we develop an accurate and robust computer model in ASPEN Plus by

implementing simulation blocks and process variables within the simulation. After much

trial and error, the successful model accurately simulates the available data representing

107
the true operational performance of the system. After establishing this working model,

we incorporate adjustments in the process parameters of the computer simulation for the

objective of process optimization. We observe the results of these process changes. For

example, these process changes and the related results could include the positive effects

of process optimization, or the negative effects to the system due to loss of process

utilities (e.g., loss of cooling water flow to a reactor).

Next, we perform sensitivity analyses to determine the dependence of output

variables on manipulated variables (e.g., the effect of column pressure on NOx

concentration leaving the top of the column). ASPEN Plus can generate plots, called

sensitivity plots, from the simulation data for a visual representation of these trends.

Finally, we can simulate various competing retrofit and process optimization options on

the computer. The model helps us assess the relative effectiveness and economic

viability of the options. We know that the model behaves as the real system does before

retrofit. We have confidence that the model accurately simulates how the system will

behave after the retrofit options and changes in process variables are implemented. In

this way, we can judge all these options based on the simulated performance of the

process.

In this case, it becomes evident early on that the conversion model is not returning

results consistent with the research we have done on how systems similar to the one at

RFAAP respond to processing changes. Therefore, we present the results in the

following sections of this chapter to prove this fact to the reader. The reader should note,

108
however, that this fact is not obvious without adequate research of the literature, because

the model is accurate for the base-case data and for some minor process changes.

However, the conversion model shows disappointing precision for changes in the most

important factors affecting the NOx abatement processes. These factors should have

yield dramatic changes in the results, but the conversion model erroneously shows them

to be minimal.

109
3.4 Discussion of the Conversion Model Results

3.4.1 The Preliminary ASPEN Model

[Link] Scrubber/Absorber

For the model, we assume that reactions (3.3) and (3.4) predominate in the

scrubber/absorber process. Through trial and error, we find fractional conversions for the

reactions for the different stoichiometric reactors as shown in Table 3.9. The overall

mass balance matches that of the data provided by RFAAP, but the set of fractional

conversion values for individual stoichiometric reactors represents only one of many

possible sets.

110
 Table 3.9. Reactions in the scrubber/absorber and their percent conversion of the

reactant shown.

Rxn # Reaction Stoichiometry Reactor Fractional Key

Name/Location Conversion Reactant

3.3 3NO2 + H2O ↔ 2HNO3 + NO ABSREACT 0.83 NO2

3.4 2NO + O2 ↔ 2NO2 ABSREACT 0.13 NO

3.3 3NO2 + H2O ↔ 2HNO3 + NO ABSORBER 0.1/tray NO2

3.4 2NO + O2 ↔ 2NO2 ABSORBER 0 NO

3.3 3NO2 + H2O ↔ 2HNO3 + NO SCRREACT 0.98 NO2

3.4 2NO + O2 ↔ 2NO2 SCRREACT 0 NO

3.3 3NO2 + H2O ↔ 2HNO3 + NO BOTREACT 0.9 NO2

3.4 2NO + O2 ↔ 2NO2 BOTREACT 0.9 NO

[Link] Demister

We represent the demister as a simple flash drum in the ASPEN model. Liquid

exiting the bottom of the demister is sent to the drain. We have assumed a negligible

amount of liquid entrainment at the top of the tower. Therefore, the ASPEN results show

no flow to the drain.

111
 [Link] Heat Exchangers and Process Heaters

We treat the preheater and the direct-fired heater as simple process stream heaters.

ASPEN calculates the enthalpy change of the process streams for the specified

temperature change. We can calculate steam or natural gas requirements from the

amount of heat transferred to the process stream. The feed/effluent heat exchanger

transfers heat from the process stream leaving the catalyst vessel (RECOUT) to the

stream leaving the preheater (VAPOR). We treat it as a heat exchanger in the ASPEN

model, specifying the outlet temperature of the hot stream (RECOUT). Table 3.10

outlines the temperature change and heat duty that the individual heating units incur on

the process streams involved.

Table 3.10. Stream temperature changes and heat duties for the heat exchangers

and process heaters.

Process Equipment ASPEN Inlet Outlet Heat Duty,

Stream Name Desig. Temp., oF Temp., oF BTU/hr

FLASHOUT Preheater PREHEAT 80 100 543563

VAPOR Economizer HEATX 100 357 1384333

RECOUT Economizer HEATX 600 350 -1384333

VAPOROUT Direct-Fired DIRHEAT 357 600 506077

Heater

112
 [Link] Catalyst Vessel

As in the scrubber/absorber, we chose fractional conversions to accurately match

the operating data for the SCR. Table 3.11 shows the important reactions and the percent

conversion of ammonia (NH3) feed.

Table 3.11. Reactions in the catalyst vessel and their percent conversion of the

reactant shown.

Rxn # Reaction Fractional Conversion Key Reactant

3.6 4NH3 + 6NO ↔ 5N2 + 6H2O 0.56 NH3

3.7 4NH3 + 4NO + O2 ↔ 4N2 + 6H2O 0.06 NH3

3.8 2NO2 + 4NH3 + O2 ↔ 3N2 + 6H2O 0.26 NH3

3.9 NO + NO2 + 2NH3 ↔ 2N2 + 3H2O 0.12 NH3

Table 3.12 gives the outputs of the system as calculated by ASPEN (simulation

results) as compared to the mass balance provided by RFAAP (RFAAP data). We show

numbers of key interest in bold. We bring the reader’s attention to the disparity between

the molar flow rate of water in the waste-acid stream (WASTACID). This disparity

results because we believe the value provided by RFAAP for the water content of this

stream to be erroneous. With 22.17 lbmol/hr (399.4 lb/hr) of filtered water fed into the

top of the absorber, for 13.1 lbmol/hr (236.1 lb/hr) to emerge at the bottom, 9.07 lbmol/hr

(163.4 lb/hr) must evaporate and be carried with the gas stream inside the

113
scrubber/absorber. The fume stream should be nearly saturated with water vapor before

it enters the absorber. In addition, the fume stream cools somewhat within the column

(from 90 oF to approximately 80 oF). Therefore, it is unlikely to cause the evaporation of

that much water.

114
 Table 3.12. Output flows as calculated by ASPEN compared to those supplied by

RFAAP.

Stream: WASTACID Acid From PRODUCT Gas to Vent

Mole Flow (Simulation Scrubber (Simulation Stack

Lbmol/hr Results) (RFAAP Data) Results) (RFAAP Data)

NO 9E-3 - 0.06 0.06

NO2 6E-3 - 9.7E-3 trace

NH3 - - 0 trace

O2 1.7E-6 - 132.1 132.06

N2 1.7E-6 - 498.8 496.8

H2O 21.4 13.1 15.9 23.9

H2SO4 - - - -

HNO2 - - - -

HNO3 2.091 2.1 - -

Total Flow 25.6 20.8 646.8 654.8

lbmol/hr

Total Flow lb/hr 556 376 18487 18631

Total Flow cuft/hr 7.8 - 3.9E5 -

Temperature F 80 86 350 350

Pressure psi 14 60 14.4 14.1

115
 Though the model accurately simulates the system, the assumptions inherent in

this configuration could create some problems in future modeling. For example, the

incorporation of multiple stoichiometric reactors may not adhere to the true mechanism

of the reactions. We simply do not know where the reactions take place. Specifically, all

the reactions could occur in the bottom of the scrubber section, or each reaction could

occur in different phases or locations in the column. Most likely, however, the reactions

occur throughout the column. The conversion model, however, gives us no information

in this area, because we have specified the reactions to take place in the sections where

the reactors are placed. This fact becomes noteworthy when considering the information

on the reactions given by the equilibrium model. We shall discuss this information and

its accuracy in the following chapter.

The assumption of stoichiometric reactors most obviously becomes a problem

when we change the number of stages in the simulation of the column. In that case,

fewer stages would definitely reduce the separation performance. Also the effective

reduction in volume of the absorber as a reaction vessel would decrease the conversion of

NOx gases to nitric acid, further increasing the NOx gases to escape out the top.

Obviously, increasing the number of stages has the opposite effect.

Modeling the absorber with reactive absorption reduces the potential for error, but

we have assumed a conversion factor on the stages that may vary with true conversion.

This problem warrants further study of the reaction mechanism for accurate modeling of

design changes. The reader should note that we neglected reactions 2 and 3 in Table 3.1

116
altogether. We choose the conversions for reaction to match the operating data for the

scrubber/absorber. With these conversions, our simulation matches the mass-balance

data given by RFAAP. Initially, the reaction conversions shown in Table 3.9 may seem

to be an arbitrary selection for reaction conversions. However, since the real process

must adhere to the conservation of mass, the overall conversion in the model matches

well with the running data provided by RFAAP. This accurate match should lend

confidence to the assumptions made above for the overall mass balance. The question

remains where the reactions tend to occur and whether these assumptions will affect

further sensitivity tests and retrofit analyses using the model. We state the potential

problems with the assumptions for the sake of completeness and for insight in the event

of further research into the absorption process.

3.4.2 Absorption-Column Performance

The performance of the NOx abatement system hinges on the performance of the

scrubber/absorber column and the catalyst vessel itself. This section of this report covers

the performance of these two key units in greater detail. The absorption column accounts

for most of the removal of the total NOx gases (on a mass basis) fed to the system. Also,

the reactions in the column set up the production of NO at the expense of NO2. Although

it possesses some potential for NO2 reduction, the catalyst vessel is designed primarily to

reduce NO with ammonia. Therefore, satisfactory performance of the catalyst vessel

hinges on the desired output from the absorption column.

117
 Figure 3.19 shows the profile of the absorption column in temperature and

pressure on each stage as calculated by ASPEN. The temperature gradient would have a

definite effect (though to an undetermined extent) on the reaction rate in treating the

reactions rigorously. We specify the pressure profile by giving a pressure drop per stage

in ASPEN. The temperature on each stage is calculated by ASPEN from the

thermodynamic model.

Figure 3.20 shows the performance of the absorption column with respect to the

reduction of NOx vapors as calculated by ASPEN. The figure clearly shows that NO is

negligibly soluble in water. NO is enriched in the vapor towards the top of the column.

The column also tends to enrich NO2 toward the top stage. However, the combined

effect of consumption of NO2 by reaction results in a net decrease in NO2 in the liquid

and vapor up the column.

118
 74 25

72
20
70
F

Stage pressure, psia

o
Stage temperature,

68
15
66
10
64

62 Temperature
5
Pressure
60

58 0
1 2 3 4 5 6 7

Theoretical stage number (1=top, 7= bottom)

Figure 3.19. Temperature and pressure profile for the simulation of the absorption tower. The abscissa corresponds to the
theoretical stages in the ASPEN model. ASPEN numbers column stages from top to bottom. Stage 1 corresponds to the top
stage (stage 1) in the real column, and stage 7 here corresponds to the bottom stage (stage 16) of the real column.

119
 0.0025

for liquid and vapor, (lbmol/hr x / lbmol/hr total)

Molar composition of NO and NO2
0.002

N O vapor N O 2 vapor
0.0015

N O liquid N O 2 liquid

0.001

0.0005

0
1 2 3 4 5 6 7
Equilibrium stage number

Figure 3.20. Absorber column profile for molar compositions of NO and NO2. Figure shows the liquid and vapor on the theoretical
absorber stages. Stage 1 is the top stage, and stage 7 is the bottom stage.

120
 Cheremisinoff and Young (1977) present work concerning scrubber performance

for the removal of NO2. They show that scrubber performance, measured by percent

removal of NO2, decreases with increasing temperature, decreasing NO2 concentration,

and increasing gas flow rate. From the data by Cheremisinoff and Young, we have

constructed a theoretical stage profile for NO2 concentration for the scrubber/absorber.

Figure 3.21 compares the data from Cheremisinoff and Young to that calculated by the

conversion-model simulation.

We clearly see that the conversion fits the general trend of the experimental data.

The conversion model qualitatively mimics the dramatic NOx removal in the scrubber

section of the scrubber/absorber tower. The removal levels off higher up the column.

However, the NO2 levels drop too steeply in the scrubber and levels off too much in the

absorption tower. This results from the arbitrary nature of the percent conversion

specifications for the reactions within the model. The data by Bowman et al. clearly

show that the driving force for NO2 removal decreases as more and more NO2 is

removed. The same trend results in the conversion model by sheer serendipity.

We choose the percent conversion in ABSREACT as 0.83 knowing that the bulk

of NO2 would likely be removed where the concentration is highest. The percent

conversion we choose for the absorption trays in ABSORBER is 0.1 per tray by reaction

(3.3). These factors are not functions of the initial NO2 concentrations as they are in

reality (reaction rates depend on concentrations). Figure 3.22 shows the problem that

121
results from the omission of this capability from the conversion model. The data by

Bowman [Link]. give a much more realistic curve. We see the overly dramatic NO2

removal of the conversion model scrubber section seen here. Again, we see the same

general trend for the conversion model as with the experimental results. However, the

NO2 removal is overly optimistic and does not change enough with respect to inlet NO2

concentration. We could possibly have lived with this lack of accuracy if the model

proved precise for conditions used for optimization. As we will show later, that is not the

case.

122
 4

3 .5
(C h e rem isin o ff an d Y o u n g, 1 9 7 7 )
NO2 on (exiting) stage, lbmol/hr 3 C o n versio n -m o d el resu lts

2 .5

1 .5

0 .5

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
S tage n u m b er
Figure 3.21. Scrubber/absrober column profile for NO2 flow rate. Stage 1 represents the top tray of the absorption column, stage
16 the last bubble-cap tray. Stage number 17 represents the upper scrubber stage (SCRUBTOP). Stage 18 the lower scrubber
stage (SCRUBBOT). Stage 19 represents the fumes fed to the column.

123
 6000

C h erem isin o ff an d Y o u n g, 1 9 7 7
5000 C o n versio n -m o d el resu lts

NO2 leaving top scrubber section, ppmv

N O 2 in le t co n cen tratio n (y=x)

4000

3000

2000

1000

0
0 1000 2000 3000 4000 5000 6000
N O 2 in let co n cen tratio n , p p m v

Figure 3.22. Comparison of conversion-model results to those presented by Cheremisinoff and Young (1977). We show the y =
x line (in = out) to simplify comparison of the two data sources. The conversion model gives optimistic results for NO2
removal. Also, the removal rate does not vary with inlet concentration to the degree that the experimental results do.

124
 3.4.3 Sensitivity Analysis

We alluded in Chapter two what the experimental literature has found to be the

most important process variables for NOx absorption and SCR. In the previous sections,

we have seen that the conversion model, though not perfect, follows the general trends

expected for the processes of NOx absorption and SCR. The question must be answered

as to whether or not we can accept the inaccuracies of the model and move on, using it as

a process optimization and prediction tool. We have alluded to the fact that we could not.

The following figures and explanation show that the conversion model cannot predict

responses to the most important process changes as compared to the experimental and

theoretical literature.

[Link] Fume-Feed Temperature

We treat the scrubber drums in ASPEN as adiabatic units. Therefore, the fume-

feed temperature determines the scrubber temperatures for the most part. Filtered water

temperature, and heats of reaction and absorption have minimal effects on temperature.

We can ignore these small effects as negligible and equate fume-feed temperature with

scrubber temperature. Reducing the NOx absorption temperature aids in absorption in

nearly all aspects. We predict that reducing the fume-feed temperature, as a means of

reducing absorption temperature, to have a dramatic effect for improving NOx

125
absorption. If the conversion model can predict this behavior, we should see reductions

in NOx escaping the top of the column, as well as an increase in nitric acid leaving the

bottom, for each successively lower temperature.

Figure 3.23 illustrates the comparison between what the literature predicts for the

effect of process temperature on NOx absorption to the results obtained from the

conversion model. Clearly, the model does not react to temperature in the manner that

the literature predicts should be the case. This follows because of the constant reaction

conversions and the equilibrium-stage calculation approach of ASPEN for the simulation.

With constant reaction conversions, temperature does not alter the ratio of reactants that

get changed to products in the stoichiometric reactors. With the equilibrium approach to

the calculation of vapor-liquid equilibrium, the rate of absorption never enters into the

calculations. Therefore, if temperature affects the equilibrium of absorption moderately,

but the rate of absorption greatly, an equilibrium model will underestimate the effect.

With the inability to predict response to temperature, we show a major shortcoming of

the conversion model.

126
 70

60

(lbmol N O 2 ,in - lbmol N O 2 ,out)/lbmol N O 2 , in

50

NO 2 re moval efficie ncy, 40

30

20
C onversion model results

10 C heremisinoff and Young, 1977

0
30 40 50 60 70 80 90 100
o
Scrubber tempe rature, F

Figure 3.23. Sensitivity plot of the effect of scrubber temperature on NO2 removal efficiency.

127
 [Link] Top-Stage Pressure

We have specified a constant pressure drop per stage for the absorption column.

Specification of the top-stage pressure of the absorption column, therefore, determines

the pressure throughout the rest of the column. Experimental results have shown that

NOx absorption is strongly dependent on the pressure of the process. Figure 3.24 shows

the effect of scrubber/absorber pressure on NO2 escaping the top of the column and

HNO3 leaving in the waste-acid stream.

[Link] Water Flow rate

Increasing the flow rate of filtered water provides the NOx with more “room” to

absorb. Also, for certain temperature, pressure, and NOx concentrations, there exists a

maximum nitric acid concentration above which no further NOx will absorb. Obviously,

running higher water flow rates dilutes the nitric acid and allows more NOx to absorb.

Again, the stoichiometric reactors do not take this phenomenon into account when

calculating the reaction conversion. Therefore, we predict increasing the filtered-water

flow rate will have little effect on the NOx removal efficiency. Figure 3.25 shows the

effect of scrubber temperature on the NO2 removal efficiency.

128
 2.2 0 .3
0 .2 8
2
0 .2 6
1.8
0 .2 4
H N O 3 in WASTAC ID , lbmol/hr

1.6 0 .2 2

N O 2 in AB SOU T, lbmol/hr
0 .2
1.4
0 .1 8
1.2 H N O3 NO2
0 .1 6

1 0 .1 4
0 .1 2
0.8
0 .1
0.6 0 .0 8

0.4 0 .0 6
0 .0 4
0.2
0 .0 2
0 0
18 19 20 21 22 22 23 24 25 26 27 28
Filtered w ater fow rate, lbmol/hr

Figure 3.24. Sensitivity plot of the effect of filtered water flow rate on NO2 and HNO3 leaving the scrubber/absorber.

129
 0 .3

2
0 .2 5

H N O 3 in WASTACID, lbmol/hr

NO 2 in ABSO UT, lbmo l/hr

1 .5 0 .2

0 .1 5
1

0 .1

0 .5 H N O3 N O2
0 .0 5

0 0
10 12 14 15 16 18 20 22 24 26 28 30
Pressu re o f abso rber to p stage, psia

Figure 3.25. Sensitivity plot of the effect of scrubber/absorber pressure on NO2 and HNO3 leaving the scrubber/absorber.

130
3.5 Conclusions

3.5.1 Conclusions Regarding Process Variables

We have seen that certain process variables strongly affect the absorption

efficiency of NOx. The conversion model corroborates the evidence presented from the

literature qualitatively in that the general trends are repeated therein. However, the

degree to which the literature predicts that the absorption should be affected does not

appear in the conversion-model simulation results. We have explained that this stems

from the constant reaction conversions of the model.

Although this conclusion represents a limitation of the conversion model, it also

gives a very important bit of information concerning NOx absorption. A revelation of

supreme importance is that reactions take in the process of NOx “absorption.” In other

words, the process of NOx removal by absorption or scrubbing must be approached as

more of a reaction engineering problem than that of simple mass transfer. Reactions

determine the total amount of NOx absorbed and, to a lesser extent, the rate at which

NOx can be absorbed.

Approaching NOx absorption as a reaction network makes clear some of the

advantages that can be gained from certain process variables that are usually ignored

when dealing with typical absorption and scrubbing processes. Temperature, for

131
example, takes a secondary role in most absorption processes. However, in the case of

NOx absorption, it has a surprisingly dramatic effect.

The domination of reaction makes inconsequential some of the key variables

usually associated with optimization of these same processes. For example, nitric acid

concentration hinders NOx absorption relatively little up to almost 35% acid by weight.

In fact, it is theorized that nitric acid can improve NOx absorption under some

circumstances. For most absorption processes, increasing the driving force for mass

transfer requires a low liquid concentration of the component being absorbed. This is not

necessarily the case in NOx absorption.

3.5.2 Problems with the First Equilibrium Model

As stated above, the main failure of the conversion model arises from the inability

of constant reaction conversions to adjust to process variables that, in reality, have

dramatic effects on reaction rates and equilibria. This situation manifests itself in

reduced responses to changes in these process variables than we observe in the actual

process. We have discussed some of the more important variables in this chapter. We

arrived at several conclusions regarding the conversion model.

1. The conversion model yields errors of almost 100% compared to literature

predictions for the ratio of NO2 fed to that removed as in Figure 3.22.

132
 2. The conversion model gives disappointing results for the effect of filtered

water flow rate on NOx removal efficiency as seen in Figure 3.23.

3. The conversion model fails to predict the effect of column pressure on NOx

removal efficiency as seen in Figure 3.23.

4. The response of NO2 absorbed in the scrubber for changes in temperature is an

order of magnitude lower than that of experimental results presented by Cheremisnoff

and Young (1977) for a similar process.

5. The response of the NOx absorption system is consistent with those displayed

by mass-transfer-only processes where reaction is not encountered.

133
3.6 Recommendations

1. Use a conversion model for creating a mass balance and for judging the effect

of process variables on the mass-transfer aspects of the process.

2. Explore the use of column-cooling techniques for new and existing NOx

absorption processes.

3. Feed NOx in as undiluted a form as is feasible.

4. Increase column pressure for increased NOx removal and viable acid recovery.

5. Increase filtered-water flow rate for increased NOx removal and decrease flow

rate for increased acid concentration.

6. Exercise strict control of ammonia injection to a selective catalytic reduction

reaction vessel.

134
3.8 Nomenclature

atm atmospheres of pressure

DEQ Department of Environmental Quality

lbs/hr pounds per hour, mass flow rate

kPa kilopascals of pressure

NRTL non-random two-liquid thermodynamic model

NC nitrocellulose

ppmv parts per million by volume

PSIA pounds per square inch (absolute) of pressure

scfm standard cubic feet per minute, volumetric flow rate

tons/yr tons per year

135
3.9 References

Cheremisinoff, P. N., and R. A. Young, R. A., Ed., Air Pollution Control and Design

Handbook, Part 2, Marcel Dekker, Inc., New York (1977), pp. 628, 671.

136